C-CAP Maui, Hawaii 2005-2010 Land Cover Change Analysis

Date that the source FGDC record was last modified.

Converted from FGDC (Federal Geographic Data Committee) CSDGM (Content Standard for Digital Geospatial Metadata) (version FGDC-STD-001-1998) using 'fgdc_to_inport_xml.pl' script. Contact Tyler Christensen (NOS) for details.

Partial upload of Positional Accuracy fields only.

Partial upload to move data access links to Distribution Info.

NOAA OCM | Source Geospatial Form: remote-sensing image | Type of Source Media: CD-ROM

Process_Description

This dataset, and the 2010-era classification was created by Photo Science. The 2005-era C-CAP classification was created by Sanborn.

This dataset contains the classifications of 2005 imagery and 2010 imagery. It is used for change analysis between these years. The

attributes include a from-to change description category. There are

522 classes corresponding to the 522 possible from-to change combinations.

In this section of the metadata, the 2005 classification

procedure is described, then the 2010 change detection classification procedure is

described. Finally in the post-processing section, the process of combining

the maps to generate a spatial matrix of change is discussed.

Early Date (2005) Classification: This section outlines the classification procedure for the Maui High Resolution C-CAP. Quickbird imagery used in producing this land cover product was also utilized in producing an associated impervious surfaces layer for the island. The mapping approach utilized a boundary summary and refinement procedure developed by Sanborn to leverage previous moderate resolution C-CAP mapping efforts to create the high resolution land cover data set. A calibration visit was conducted and training data was collected for classification. Non impervious features were mapped using a 0.5 acre minimum mapping unit (MMU).

Pre-processing steps:

The Quickbird mosaic utilized for this project was provided from Digital Globe as an orthorectified, georeferenced product and was consistent with the quoted 1:12000 spatial accuracy. In conjunction with the Quickbird mosaic, Sanborn used an air photo mosaic from the USGS for the island collected in 2004 as well as a DTM created from a 5m radar collect for the island. The original return image from the radar was also utilized.

Field-Collected Data:

Training data was collected in the field between March 30th and April 11th, 2008. Land cover observations were recorded by analysts. The referenced dataset went through a QA/QC procedure to ensure it met accuracy standards.

Classification:

Initial Segmentation - Image segmentation was completed at multiple scales using the multispectral (2.4 m) imagery in order to group like spectral and textural objects within the imagery. The initial segments were created at a larger scale (scale refers to the target size and shape of like features generalized by Definiens software) of 100. Larger segments are required to leverage the 30 m data sets and incorporate the IFSAR data that is collected at spatial resolution of 5 m. For consistency, the associated impervious data set was incorporated into the segmentation layer as a boundary delimiter. Segments can share boundaries of an impervious surface, but can never overlap an impervious surface.

Rule set Creation - The initial map of the island of Maui is created by incorporating an existing C-CAP 2001 land cover data set through a logical rule set created by a Sanborn analyst. The rules are created such that image segments are labeled to the C-CAP classification scheme based on their underlying medium resolution components.

Initial Label and Edits - The automated labeling procedure created a reasonable representation of land cover using the labeling routine, however errors did occur. Gross inaccuracies in the data product that would inhibit the further classification of the segments at a smaller scale factor were corrected through knowledge based models or manual edits.

Secondary Segmentation - Further refinement of the map is only possible though classification of the image segments directly. Classification of the image segments is a hierarchical process, whereby larger scale segments are composed of smaller scale segments. The initial segmentation was used to further aid in the automated classification of the secondary (more detailed) image segments. The final scale factor for the mosaic was 35.

Automated Classification - Automated classification of the image segments is accomplished through a process modeling incorporated within Definiens software. Models are built to refine or reclassify land cover areas by utilizing the wealth of attribute information linked to each segment within Definiens. In a small amount of cases, the nearest neighbor classification routine was used to refine segments. Nearest neighbor training data is collected through analyst interpretation or ancillary data sets and used to calibrate the classification algorithm.

Automated Classification Edits - As with any automated or semi-automated land cover classification there are often inconsistencies in the land cover map. The final step before map finalization was to remove inaccuracies through manual segment labeling as interpreted by an analyst.

Map Finalization - Sanborn used independent reviewer's comments to further refine the land cover map.

Attributes for this product are as follows:

0 Background

1 Unclassified

2 Impervious

3

4

5 Developed, Open Space

6 Cultivated Crops

7 Pasture/Hay

8 Grassland/Herbaceous

9 Deciduous Forest

10 Evergreen Forest

11 Mixed Forest

12 Scrub/Shrub

13 Palustrine Forested Wetland

14 Palustrine Scrub/Shrub Wetland

15 Palustrine Emergent Wetland

16 Estuarine Forested Wetland

17 Estuarine Scrub/Shrub Wetland

18 Estuarine Emergent Wetland

19 Unconsolidated Shore

20 Bare Land

21 Open Water

22 Palustrine Aquatic Bed

23 Estuarine Aquatic Bed

24 Tundra

25 Snow/Ice

Late Date (2010) Classification:

This section outlines the classification procedure for the Maui High Resolution C-CAP. WorldView2 imagery used in producing this land cover product was also utilized in producing an associated impervious surfaces layer for the island. The mapping approach utilized a boundary summary, outlier change detection process and refinement procedure developed by Photo Science to leverage the 2005 high resolution C-CAP data set to create the 2010 high resolution C-CAP. A calibration visit was not conducted though NOAA had access to local resources for validation. Non impervious features were mapped using a 0.25 acre minimum mapping unit (MMU) and impervious features were mapped using a 0.1 acre MMU.

Pre-processing steps:

The WorldView2 mosaic utilized for this project was provided from Digital Globe as an orthorectified, georeferenced product. Multiple image primitives and indicies such as texture and NDVI were derived from the 8-band satellite data. The imagery was re-sampled from it's native 2m spatial resolution to 2.4m. The 2005 C-CAP classification and the base Quickbird imagery (used in the original mapping) were geometrically corrected to co-register to the WorldView2 data which has a higher geolocation accuracy.

Impervious Update:

The 2005 impervious was over-laid with semi-transparency on the 2010 satellite data. It was panned at a scale of 1:3,500 and manually updated to match the 2010 data. Analysts zoomed in to a larger scale when necessary to perform edits. The product went through a QC procedure to ensure features were accurately captured.

Segmentation and Outlier Detection Process:

The 2010 impervious was combined with the 2005 land cover to create a hybrid data set. Image segmentation, done in Trimble's eCognition software, was completed at multiple scales using the multispectral (2.4 m) imagery in order to group like spectral and textural objects within the imagery. For consistency, the associated hybrid data set was incorporated into the segmentation layer as a boundary delimiter. Segments contained image attributes and a label from the 2005 classification. These data were inputs to a custom multi-variate outlier detection tool that identified objects of potential change. These areas created the change mask.

Training Site Data:

Training data was photo interpreted and collected within the change mask and was super-sampled from features outside of the mask. The referenced data set went through a QA/QC procedure to ensure it met accuracy standards.

Classification:

Automated Classification - Image segments were classified using a decision tree classifier in Rulequest's See5 software based on the training data and image attributes.

Automated Classification Refinement - Models are built to refine or reclassify land cover areas by utilizing the wealth of attribute information linked to each segment within eCognition.

Classification Edits - As with any automated or semi-automated land cover classification there are often inconsistencies in the land cover map. The final step before map finalization was to remove inaccuracies through manual segment labeling as interpreted by an analyst.

Map Finalization - Photo Science used independent reviewer's comments to further refine the land cover map.

Attributes for this product are as follows:

0 Background

1 Unclassified

2 Impervious

3

4

5 Developed, Open Space

6 Cultivated Crops

7 Pasture/Hay

8 Grassland/Herbaceous

9 Deciduous Forest

10 Evergreen Forest

11 Mixed Forest

12 Scrub/Shrub

13 Palustrine Forested Wetland

14 Palustrine Scrub/Shrub Wetland

15 Palustrine Emergent Wetland

16 Estuarine Forested Wetland

17 Estuarine Scrub/Shrub Wetland

18 Estuarine Emergent Wetland

19 Unconsolidated Shore

20 Bare Land

21 Open Water

22 Palustrine Aquatic Bed

Post-Processing Steps:

A GIS matrix algorithm was run with 2005 and 2010 datasets as inputs. The algorithm generates

a from-to file where each combination is uniquely labeled from 1 to 625.

Metadata imported